[PDF][PDF] Contrastive learning of global and local features for medical image segmentation with limited annotations
E Konukoglu - proceedings.neurips.cc
We use a UNet [7] based encoder (e)-decoder (d) architecture. e consists of 6 convolutional
blocks, each consisting of two 3× 3 convolutions followed by a 2× 2 maxpooling layer with …
blocks, each consisting of two 3× 3 convolutions followed by a 2× 2 maxpooling layer with …
[PDF][PDF] Contrastive learning of global and local features for medical image segmentation with limited annotations
KCEEN Karani, E Konukoglu - arXiv preprint arXiv:2006.10511, 2020 - researchgate.net
A key requirement for the success of supervised deep learning is a large labeled dataset-a
condition that is difficult to meet in medical image analysis. Selfsupervised learning (SSL) …
condition that is difficult to meet in medical image analysis. Selfsupervised learning (SSL) …